Datasets
The README links benchmark and dataset catalogues rather than fixing one data release.
Benchmark
Open Problems
Open Problems is an extensible platform hosting benchmark tasks and their datasets.
Sources
openproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource linksOverview
Metrics
No platform-wide scientific metric applies: hosted task definitions select their own evaluator.
inapplicable
Allowed inputs
Inputs are specified by each hosted task.
Sources
openproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource linksEvaluation procedure diagram
How it worksEvaluation procedure
Conceptual procedure. Task variants and protocol versions retain their separate scoring conditions.
Sources
openproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource linkslimited source coverage · Automated source review, 2026-09-16. All specifications and missing details
Results
Each comparison retains its reviewed evaluation scope, dataset and metric. Results are shown without a pooled ranking.
Open Problems label projection CENGEN-BATCH-ACCURACY: Label projection on CeNGEN (split by batch), Accuracy
accuracy (fraction) · Higher values are better.
Open Problems label projection CENGEN-BATCH-ACCURACY: Label projection on CeNGEN (split by batch), Accuracy · CeNGEN (split by batch) (Open Problems label projection split)
Evidence origin: Author-reported evaluation.
openproblems-label primary benchmark evidence · results, dataset(cengen_batch), metric(accuracy)- Results published by the Open Problems project, source checked but not independently reproduced.
- This covers the label projection task at v1.0.0 only, not the whole Open Problems suite.
- true_labels and random_labels are controls that bound the scale, not competing methods.
- Preprocessing is part of the run, so the same method appears once per parameter set.
Comparison details and limitations
Every method Open Problems label projection reports on Label projection on CeNGEN (split by batch), Accuracy, scored with Accuracy on CeNGEN (split by batch).
Automated source review: 2026-09-18. Numerical source review does not establish independent reproduction.
Dots show point estimates. Whiskers show only explicitly defined uncertainty (standard deviation, standard error or a labelled interval); their definitions remain in Table. Unresolved uncertainty is not plotted. Differences do not establish statistical significance.
Showing 12 of 16 matching rows.
Tested configuration
00.250.50.751
Reported score- True Labels1
- Logistic regression (log CP10k)0.878
- Multilayer perceptron (log scran)0.857
- Seurat reference mapping (SCTransform)0.848
- XGBoost (log scran)0.83
- XGBoost (log CP10k)0.827
- Multilayer perceptron (log CP10k)0.825
- K-neighbors classifier (log CP10k)0.812
- K-neighbors classifier (log scran)0.787
- Logistic regression (log scran)0.772
- scANVI (All genes)0.647
- scANVI (Seurat v3 2000 HVG)0.262
Methods and evaluation design
Procedure, tasks and evaluated configurations
How it works
Evaluation methodology
Open Problems is a collection of separately versioned single-cell evaluation tasks. Each task defines its inputs, reference data, methods, controls and metrics. For example, label projection learns from reference labels and predicts a held-out dataset, whereas integration measures how supplied batches are combined.
Sources
openproblems-label primary benchmark evidence · Label Projection v1.0.0: task description, dataset variants and controlsEvaluation design
Benchmarks bring together tasks and protocols. A task describes the biological question; a protocol defines a particular test.
Tasks
- Batch integration
- Open Problems label projection CENGEN-BATCH-ACCURACY: Label projection on CeNGEN (split by batch), Accuracy
- Open Problems label projection CENGEN-BATCH-F1: Label projection on CeNGEN (split by batch), F1 score
- Open Problems label projection CENGEN-BATCH-F1-MACRO: Label projection on CeNGEN (split by batch), Macro F1 score
- Open Problems label projection CENGEN-RANDOM-ACCURACY: Label projection on CeNGEN (random split), Accuracy
- Open Problems label projection CENGEN-RANDOM-F1: Label projection on CeNGEN (random split), F1 score
- Open Problems label projection CENGEN-RANDOM-F1-MACRO: Label projection on CeNGEN (random split), Macro F1 score
- Open Problems label projection PANCREAS-BATCH-ACCURACY: Label projection on Pancreas (by batch), Accuracy
- Open Problems label projection PANCREAS-BATCH-F1: Label projection on Pancreas (by batch), F1 score
- Open Problems label projection PANCREAS-BATCH-F1-MACRO: Label projection on Pancreas (by batch), Macro F1 score
- Open Problems label projection PANCREAS-RANDOM-ACCURACY: Label projection on Pancreas (random split), Accuracy
- Open Problems label projection PANCREAS-RANDOM-F1: Label projection on Pancreas (random split), F1 score
- Open Problems label projection PANCREAS-RANDOM-F1-MACRO: Label projection on Pancreas (random split), Macro F1 score
- Open Problems label projection PANCREAS-RANDOM-LABEL-NOISE-ACCURACY: Label projection on Pancreas (random split with label noise), Accuracy
- Open Problems label projection PANCREAS-RANDOM-LABEL-NOISE-F1: Label projection on Pancreas (random split with label noise), F1 score
- Open Problems label projection PANCREAS-RANDOM-LABEL-NOISE-F1-MACRO: Label projection on Pancreas (random split with label noise), Macro F1 score
- Open Problems label projection TABULA-MURIS-SENIS-LUNG-RANDOM-ACCURACY: Label projection on Tabula Muris Senis Lung (random split), Accuracy
- Open Problems label projection TABULA-MURIS-SENIS-LUNG-RANDOM-F1: Label projection on Tabula Muris Senis Lung (random split), F1 score
- Open Problems label projection TABULA-MURIS-SENIS-LUNG-RANDOM-F1-MACRO: Label projection on Tabula Muris Senis Lung (random split), Macro F1 score
- Open Problems label projection ZEBRAFISH-LABS-ACCURACY: Label projection on Zebrafish (by laboratory), Accuracy
- Open Problems label projection ZEBRAFISH-LABS-F1: Label projection on Zebrafish (by laboratory), F1 score
- Open Problems label projection ZEBRAFISH-LABS-F1-MACRO: Label projection on Zebrafish (by laboratory), Macro F1 score
- Open Problems label projection ZEBRAFISH-RANDOM-ACCURACY: Label projection on Zebrafish (random split), Accuracy
- Open Problems label projection ZEBRAFISH-RANDOM-F1: Label projection on Zebrafish (random split), F1 score
- Open Problems label projection ZEBRAFISH-RANDOM-F1-MACRO: Label projection on Zebrafish (random split), Macro F1 score
These source-backed links do not make different protocols or scores interchangeable.
Baseline coverage
Reference methods help show what a model adds beyond simple controls. We track a null control and a conventional method for each protocol.
No concrete protocols are explicitly linked to this suite. Protocol identification and baseline selection are outstanding.
Protocol coverage CSV · Model evaluation matrix · Source table · Release and checksums
Coverage is derived from release 2026-09-29-06401fd5b220. Source citations describe the original records; they do not validate an unreviewed baseline proposal. No results have been generated by this audit.
Run this benchmark
Open Problems links task-specific repositories and contribution documentation. The inspected label-projection repository specifies AnnData inputs, prediction outputs and evaluation components, but these READMEs do not supply a complete end-to-end shell run. Select a released task workflow, executor and resource bundle before running; the platform itself has no universal single command.
A maintained rewire runner has not been verified for this benchmark. Check data access, weights, licences, dependencies and hardware in the linked official documentation; requirements have not been fully extracted.
openproblems-bio/openproblems / README.md; openproblems-bio/task_label_projection / README.md · Platform README.md lines 9–18; task_label_projection/README.md Description and API sectionsStrengths, limitations and unresolved questions
Strengths and limitations
Strengths and considerations
- The platform separates task definitions from methods submitted to them.
Sources
openproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links
Limitations and conditions
- Results from different Open Problems tasks or split variants are not interchangeable. Perfect-label controls and random-label controls calibrate particular metrics; they are not measured biological performance ceilings.
Sources
openproblems-label primary benchmark evidence · Label Projection v1.0.0: task description, dataset variants and controls
Profile review details
Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction.
Stable record: discovery-benchmark-open-problemsSpecifications
Inputs, training, access and other details
Explanatory profile: limited source coverage · Automated source review, 2026-09-16. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.
| Property | Description and evidence |
|---|---|
| Datasets | The README links benchmark and dataset catalogues rather than fixing one data release.Sourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
| Splits | No platform-wide split exists: select a hosted benchmark task and its dataset protocol. · Not applicableSourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
| Metrics | No platform-wide scientific metric applies: hosted task definitions select their own evaluator. · Not applicableSourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
| Baselines | No single platform-wide baseline: methods and references belong to each hosted task. · Not applicableSourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
| Leakage controls | Controls are task dependent. Label Projection v1.0.0 distinguishes training/test batches and includes both random and batch-based CeNGEN dataset variants. Other tasks, such as batch integration, evaluate transductive processing of the supplied cells; a universal held-out-cell rule would be misleading.Sourcesopenproblems-label primary benchmark evidence · Label Projection v1.0.0: task description, dataset variants and controls |
| Uncertainty | The inspected task pages and reporting configuration do not prescribe one platform-wide bootstrap or repeated-seed interval. Individual task versions define datasets, metrics and runs; model uncertainty from a method such as scANVI is not benchmark-score uncertainty. · Not reported in inspected sourcesSourcesopenproblems-label primary benchmark evidence · Label Projection v1.0.0: task description, dataset variants and controls |
| Entity type | Platform hosting computational biology benchmark tasks.Sourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
| Organisms | Organisms belong to the selected task and dataset; the platform defines no unique organism. · Not applicableSourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
| Assays | The platform does not prescribe one assay. · Not applicableSourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
| Allowed inputs | Inputs are specified by each hosted task.Sourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
| Adaptation | Adaptation rules belong to the task implementation; the platform is not a fitted predictor. · Not applicableSourcesopenproblems-bio/openproblems official source · Pinned README: platform description and benchmark/dataset resource links |
Evidence
Source checking verifies the cited claim or transcription. It does not establish independent reproduction.
Papers and result coverage
Last literature check: 2026-09-17. Primary-paper discovery and source inspection. Source-checked results are not independently reproduced experiments.
| Paper or primary resource | Version | Reference |
|---|---|---|
| Defining and benchmarking open problems in single-cell analysis | PMC full-text XML snapshot PMC11030530 at retrieval; byte-pinned by SHA-256 | Read source DOI: 10.21203/rs.3.rs-4181617/v1 |
| openproblems-label primary benchmark evidence | v1.0.0 | Read source |
Historical gaps recorded on 2026-09-17
The catalogue now holds 384 result rows for this benchmark. A note below about pending extraction describes the state on 2026-09-17 and may since have been answered by a later batch. The result rows and their sources are the current record.
- Task release, dataset, preprocessing and metric must be fixed before charting.
- The page lists multiple releases; a downloaded release artifact is needed to pin exact numerical leaderboard observations.
- No values estimated from chart coordinates.
Search and extraction details
primary protocol screened
Searches
- Open Problems single cell label projection benchmark v1.0.0 paper
- CAPRI assessment rounds 46 54 protein docking results
- GEARS predicting transcriptional outcomes multigene perturbations 2023
- scVI scANVI reference mapping benchmark primary paper
Evidence locations
- Primary paper: Label projection
- Versioned label projection page: task definition, normalization, leaderboard
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
19 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Diagram caption Conceptual procedure. Task variants and protocol versions retain their separate scoring conditions. Individual claims | openproblems-bio/openproblems official source Pinned README: platform description and benchmark/dataset resource links Version: 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131 | source checked automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Diagram steps
| openproblems-bio/openproblems official source Pinned README: platform description and benchmark/dataset resource links Version: 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131 | source checked automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Diagram title Evaluation procedure Individual claims | openproblems-bio/openproblems official source Pinned README: platform description and benchmark/dataset resource links Version: 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131 | source checked automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Datasets The README links benchmark and dataset catalogues rather than fixing one data release. Individual claims | openproblems-bio/openproblems official source Pinned README: platform description and benchmark/dataset resource links Version: 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131 | source checked automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Splits No platform-wide split exists: select a hosted benchmark task and its dataset protocol. Individual claims | openproblems-bio/openproblems official source Pinned README: platform description and benchmark/dataset resource links Version: 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131 | inapplicable automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Adaptation Adaptation rules belong to the task implementation; the platform is not a fitted predictor. Individual claims | openproblems-bio/openproblems official source Pinned README: platform description and benchmark/dataset resource links Version: 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131 | inapplicable automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Metrics No platform-wide scientific metric applies: hosted task definitions select their own evaluator. Individual claims | openproblems-bio/openproblems official source Pinned README: platform description and benchmark/dataset resource links Version: 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131 | inapplicable automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Baselines No single platform-wide baseline: methods and references belong to each hosted task. Individual claims | openproblems-bio/openproblems official source Pinned README: platform description and benchmark/dataset resource links Version: 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131 | inapplicable automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Leakage controls Controls are task dependent. Label Projection v1.0.0 distinguishes training/test batches and includes both random and batch-based CeNGEN dataset variants. Other tasks, such as batch integration, evaluate transductive processing of the supplied cells; a universal held-out-cell rule would be misleading. Individual claims | openproblems-label primary benchmark evidence Label Projection v1.0.0: task description, dataset variants and controls Version: v1.0.0 | source checked automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Uncertainty The inspected task pages and reporting configuration do not prescribe one platform-wide bootstrap or repeated-seed interval. Individual task versions define datasets, metrics and runs; model uncertainty from a method such as scANVI is not benchmark-score uncertainty. Individual claims | openproblems-label primary benchmark evidence Label Projection v1.0.0: task description, dataset variants and controls Version: v1.0.0 | unreported automated source review · 2026-09-16 Audit detailsPrimary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Sources and history
View linked audit checks and correction history
Release 2026-09-29-06401fd5b220 · Record review: discovered
6 source records and release history
- openproblems-bio/openproblems official source · Original source · 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131
- openproblems-label primary benchmark evidence · Original source · v1.0.0
- Defining and benchmarking open problems in single-cell analysis · Original source · PMC full-text XML snapshot PMC11030530 at retrieval; byte-pinned by SHA-256
- openproblems-bio/openproblems / README.md · Original source · Git commit 0ca5d0cd040b741c1b6cc2e4cad7230cb2c50131
- openproblems-bio/task_label_projection / README.md · Original source · Git commit e14ce24c4af90c1f7311a52db4f9e720d97cc91e
- openproblems-label primary benchmark evidence · Original source · v1.0.0
Technical metadata and extraction receipts
Stable ID: discovery-benchmark-open-problems
- areas
- single-cell
- entity level
- suite
- scope note
- Specialist molecular or omics evaluation; protocol details require review before numerical comparison.
- task
- Community single-cell analysis benchmarks
- version
- Not reported
- benchmark research
- review date: 2026-09-17; status: primary_protocol_screened; primary sources: expansion-p3-open-problems-paper; expansion-p3-open-problems; inspected locators: Primary paper: Label projection; Versioned label projection page: task definition, normalization, leaderboard; searched queries: Open Problems single cell label projection benchmark v1.0.0 paper; CAPRI assessment rounds 46 54 protein docking results; GEARS predicting transcriptional outcomes multigene perturbations 2023; scVI scANVI reference mapping benchmark primary paper; gaps: Task release, dataset, preprocessing and metric must be fixed before charting.; The page lists multiple releases; a downloaded release artifact is needed to pin exact numerical leaderboard observations.; No values estimated from chart coordinates.; claim scope: Primary-paper discovery and source inspection. Source-checked results are not independently reproduced experiments.
- historical missing metadata
- dataset release: unextracted; metric implementation: unextracted; split manifest: unextracted; version: unextracted
- metadata review scope
- historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
- entity classification
- review date: 2026-09-17; rationale: The cited profile describes a collection of evaluation tasks or protocols; retain it as the top-level benchmark suite. Its datasets and individual protocols remain separate records.; source ids: src-discovery-openproblems-bio-openproblems; source locator: Pinned README: platform description and benchmark/dataset resource links; ambiguities: None recorded
- run documentation
- record id: discovery-benchmark-open-problems; source ids: run-doc-open-problems-readme-md-0ca5d0cd; run-doc-openproblems-label-readme-md-e14ce24c; status: official_documentation_linked; summary: Open Problems links task-specific repositories and contribution documentation. The inspected label-projection repository specifies AnnData inputs, prediction outputs and evaluation components, but these READMEs do not supply a complete end-to-end shell run. Select a released task workflow, executor and resource bundle before running; the platform itself has no universal single command.; source locator: Platform README.md lines 9–18; task_label_projection/README.md Description and API sections
Related records
- evaluates task: Batch integration
- part of: Open Problems label projection CENGEN-BATCH-ACCURACY: Label projection on CeNGEN (split by batch), Accuracy
- part of: Open Problems label projection CENGEN-BATCH-F1: Label projection on CeNGEN (split by batch), F1 score
- part of: Open Problems label projection CENGEN-BATCH-F1-MACRO: Label projection on CeNGEN (split by batch), Macro F1 score
- part of: Open Problems label projection CENGEN-RANDOM-ACCURACY: Label projection on CeNGEN (random split), Accuracy
- part of: Open Problems label projection CENGEN-RANDOM-F1: Label projection on CeNGEN (random split), F1 score
- part of: Open Problems label projection CENGEN-RANDOM-F1-MACRO: Label projection on CeNGEN (random split), Macro F1 score
- part of: Open Problems label projection PANCREAS-BATCH-ACCURACY: Label projection on Pancreas (by batch), Accuracy
- part of: Open Problems label projection PANCREAS-BATCH-F1: Label projection on Pancreas (by batch), F1 score
- part of: Open Problems label projection PANCREAS-BATCH-F1-MACRO: Label projection on Pancreas (by batch), Macro F1 score
- part of: Open Problems label projection PANCREAS-RANDOM-ACCURACY: Label projection on Pancreas (random split), Accuracy
- part of: Open Problems label projection PANCREAS-RANDOM-F1: Label projection on Pancreas (random split), F1 score
- part of: Open Problems label projection PANCREAS-RANDOM-F1-MACRO: Label projection on Pancreas (random split), Macro F1 score
- part of: Open Problems label projection PANCREAS-RANDOM-LABEL-NOISE-ACCURACY: Label projection on Pancreas (random split with label noise), Accuracy
- part of: Open Problems label projection PANCREAS-RANDOM-LABEL-NOISE-F1: Label projection on Pancreas (random split with label noise), F1 score
- part of: Open Problems label projection PANCREAS-RANDOM-LABEL-NOISE-F1-MACRO: Label projection on Pancreas (random split with label noise), Macro F1 score
- part of: Open Problems label projection TABULA-MURIS-SENIS-LUNG-RANDOM-ACCURACY: Label projection on Tabula Muris Senis Lung (random split), Accuracy
- part of: Open Problems label projection TABULA-MURIS-SENIS-LUNG-RANDOM-F1: Label projection on Tabula Muris Senis Lung (random split), F1 score
- part of: Open Problems label projection TABULA-MURIS-SENIS-LUNG-RANDOM-F1-MACRO: Label projection on Tabula Muris Senis Lung (random split), Macro F1 score
- part of: Open Problems label projection ZEBRAFISH-LABS-ACCURACY: Label projection on Zebrafish (by laboratory), Accuracy
- part of: Open Problems label projection ZEBRAFISH-LABS-F1: Label projection on Zebrafish (by laboratory), F1 score
- part of: Open Problems label projection ZEBRAFISH-LABS-F1-MACRO: Label projection on Zebrafish (by laboratory), Macro F1 score
- part of: Open Problems label projection ZEBRAFISH-RANDOM-ACCURACY: Label projection on Zebrafish (random split), Accuracy
- part of: Open Problems label projection ZEBRAFISH-RANDOM-F1: Label projection on Zebrafish (random split), F1 score
- part of: Open Problems label projection ZEBRAFISH-RANDOM-F1-MACRO: Label projection on Zebrafish (random split), Macro F1 score
- subject: Open Problems: evaluates task Batch integration